A tumor microenvironment-responsive micelle co-delivered radiosensitizer Dbait and doxorubicin for the collaborative chemo-radiotherapy of glioblastoma

Abstract Glioblastoma is rather recalcitrant to existing therapies and effective interventions are needed. Here we report a novel microenvironment-responsive micellar system (ch-K5(s-s)R8-An) for the co-delivery of the radiosensitizer Dbait and the chemotherapeutic doxorubicin (DOX) to glioblastoma. Accordingly, the ch-K5(s-s)R8-An/(Dbait-DOX) micelles plus radiotherapy (RT) treatment resulted in a high degree of apoptosis and DNA damage, which significantly reduced cell viability and proliferation capacity of U251 cells to 64.0% and 16.3%, respectively. The angiopep-2-modified micelles exhibited substantial accumulation in brain-localized U251 glioblastoma xenografts in mice compared to angiopep-2-lacking micelles. The ch-K5(s-s)R8-An/(Dbait-DOX) + RT treatment group exhibited the smallest tumor size and most profound tumor tissue injury in orthotopic U251 tumors, leading to an increase in median survival time of U251 tumor-bearing mice from 26 days to 56 days. The ch-K5(s-s)R8-An/(Dbait-DOX) micelles can be targeted to brain-localized U251 tumor xenografts and sensitize the tumor to chemotherapy and radiotherapy, thereby overcoming the inherent therapeutic challenges associated with malignant glioblastoma.


Introduction
In adults, glioblastoma is the most common primary malignant brain tumor and its prognosis is particularly poor (Ostrom et al., 2014). Standard treatment includes surgery, radiotherapy (RT ), and/or chemotherapy. However, therapeutic efficacy against glioblastoma has remained poor and the malignancy typically recurs within 7 months (Siegel et al., 2020), rendering glioblastoma a life-threatening problem even in patients whose treatment was initially successful (Siegel et al., 2020). The most pivotal problems of glioblastoma treatment are inherent RT resistance (Dutreix et al., 2010;Zhang et al., 2019), tolerance to chemotherapy drugs (Fan et al., 2014;Le Rhun et al., 2019), and low drug permeability due to the blood-brain barrier and the blood-brain tumor barrier (Zhao et al., 2017;Arvanitis et al., 2020).
These obstacles and the severe adverse effects associated with conventional glioblastoma therapies have created a medical need to develop new, safer, and more effective interventions for glioblastoma (Tang et al., 2019). To improve therapeutic outcomes, combination regimens such as concurrent chemo-radiotherapy have gained traction in basic research and clinical trials (Neoptolemos et al., 2004;Song et al., 2017), addressing both the primary tumor and the metastatic lesions outside of the irradiation zone. The synergistic effects of combination therapies have been demonstrated for glioblastoma and other cancer types (Gomez et al., 2016;Zhang et al., 2017).
The treatment efficacy of chemo-radiotherapy can be expanded by pharmacologically rendering tumor tissue more susceptible to RT. To that end, stabilized DNA molecules (Dbait) were designed to mimic DNA double-strand breaks and give off a 'false' DNA damage signal, ultimately inhibiting the recruitment of proteins involved in double-strand break repair and forcing cells into apoptosis (Quanz et al., 2009;Biau et al., 2014). Recent reports have shown that the use of Dbait as radiosensitizers in combination with RT is effective for the treatment of glioblastoma, head and neck squamous cell carcinoma, skin melanoma, and colorectal cancer metastases, amongst others Biau et al., 2016;Liu et al., 2017).
We previously developed a blood-brain barrier-targeting and glioblastoma cell-targeting micellar system (ch-K5(s-s) R8-An) to deliver Dbait into glioblastoma cells and make these cells more sensitive to RT (Jiao et al., 2019). In this study, Dbait and the chemotherapeutic doxorubicin (DOX) were further co-encapsulated into ch-K5(s-s)R8-An micelles on the basis of our previous work, which aimed to mount a blitz attack on the tumor cell's DNA infrastructure by damaging DNA with RT and concomitantly ceasing macromolecular biosynthesis and topoisomerase II progression with DOX (Pommier et al., 2010;Tacar et al., 2013). Such synergistic mechanism induced maximum apoptotic signaling and consequent cell death in cultured human malignant glioblastoma (U251) cells and in situ glioblastoma xenografts in mice.
Step 1, the angiopep-2 (An) that is conjugated to the micelle outer shell serves to target the micelles to the blood-brain barrier (Huang et al., 2013;Ruan et al., 2015;Feng et al., 2017).
Step 2, the enzymatically cleavable linkers (matrix metalloproteinase 2 [MMP-2] responsive peptides) that are attached to the cell penetration-enhancing R8 peptide in the middle shell ensure the glioblastoma cell internalization following MMP-2 cleavage in the tumor microenvironment (Cui et al., 2016).
Step 3, the K5-cholesterol anchors comprise the inner shell and are cross-linked via disulfide bridges for high drug loading capacity and triggered drug release under the reductive environment in the cytosol of tumor cells . The collaborative anti-glioblastoma effect of ch-K5(s-s)R8-An/(Dbait-DOX) plus RT was investigated in vitro and in vivo.

Materials
The cholesterol-KKKKKRRRRRRRR (ch-K5R8) peptides and cholesterol-KKKKKRRRRRRRR-Angiopep-2 (ch-K5R8-An) peptides were designed by our group and then synthesized by Ontores Biotechnologies (Shanghai, China). Other materials used in this study are listed in the Supplementary Material.

In vitro cytotoxicity and anti-proliferation effect
The cytotoxicity and half maximal inhibitory concentration (IC 50 ) of the combined treatment with ch-K5(s-s)R8-An/ (Dbait-DOX) and RT were assessed with Cell Counting Kit-8 (CCK-8) assays in U251 cells. Colony formation assays were performed to determine the anti-proliferation effect of each treatment. Plating efficiency was used to evaluate the colony formation ability of each group, which was calculated by dividing the colony count by the plated cell count. See Supplementary Material for details.

In vitro apoptosis assay
The apoptosis rate of the combined treatment with ch-K5(s-s) R8-An/(Dbait-DOX) and RT was determined by flow cytometry. After treatment, U251 cells were stained by an annexin v-APC/propidium iodide (PI) apoptosis detection kit and analyzed by flow cytometry. Data were processed using FlowJo software. See Supplementary Material for details.

In vitro DNA damage and repair
Western blot assays were performed to determine DNA damage and repair in U251 cells after each treatment through the extent of phosphorylation of histone H2A (γ-H 2 AX), phospho-P53 (p-P53), and DNA-dependent protein kinase catalytic subunit (DNA-PKcs). See Supplementary Material for details.

In vivo brain targeting and biodistribution
Near-infrared fluorescent boron dipyrromethene (BODIPY) was used to investigate the in vivo brain targeting ability and biodistribution of the ch-K5(s-s)R8-An micelles. In situ U251 tumor-bearing models were constructed by implanting U251 cells into the brain tissue of nude mice. Then, U251 tumor-bearing mice were randomly divided into three groups and were injected with free BODIPY, ch-K5(s-s)R8/BODIPY micelles, or ch-K5(s-s)R8-An/BODIPY through their tail veins. After injection, biodistribution was evaluated by in vivo fluorescence imaging and the brain tumor-targeting abilities were assessed by imaging the frozen sections of brain tissues using confocal laser scanning microscopy. See Supplementary Material for details.
Twenty-four hours after the last RT treatment, intracranial tumors were imaged by magnetic resonance imaging (MRI, Magnetom Aera, Siemens, Munich, Germany) on day 28 after U251 cell implantation. Then, part of the animals were sacrificed (n = 6/group) to collect major organs (heart, liver, spleen, lung, kidney, brain), and the remaining animals were kept alive under standard care conditions. The body weight of mice was recorded every two days and survival was monitored to construct Kaplan-Meier curves (n = 6/group). The histological damage levels of the brain tumor sections were assessed by hematoxylin and eosin (H&E) staining and γ-H 2 AX immunofluorescence staining. Meanwhile, in vivo safety of each treatment was evaluated by H&E histological staining of organ sections. See Supplementary Material for details.

Statistical analyses
All data are presented as means ± standard deviation (SD). Statistical analysis was performed using one-way analysis of variance (ANOvA). The difference was considered significant when the P-value was less than 0.05.

In vitro drug release from ch-K5(s-s)R8-an/(Dbait-DOX) micelles
Dbait is a DNA double-strand break mimetic and induces hyperactivation of DNA-dependent protein kinases (DNA-PKs) involved in DNA damage repair. Accordingly, Dbait must enter the nucleus to initiate its radiosensitization effect . The release profiles of Dbait and DOX from ch-K5(s-s) R8-An/(Dbait-DOX) micelles were investigated under a pH-neutral (pH = 7.4), pH-acidic (pH = 5.5), and reductive acidic (pH = 5.5 + 10 mM DTT) environment at 37 °C in order to simulate physiological conditions in the circulation and the lysosomal compartment in tumor cells , respectively. The cytosolic glutathione (GSH) concentration in tumor cells is 100-1000 times higher than that in normal cells, which includes glioblastoma (Zheng et al., 2019), thus DTT was used to further mimic the reductive state in cancer cells normally imparted by GSH (Xiong et al., 2018).

In vitro micelle uptake and intracellular localization
The cellular uptake and intracellular localization of ch-K5(s-s) R8-An/(Dbait-DOX) were analyzed by imaging YOYO-1-labeled Dbait (red) and DOX auto fluorescence (green) by confocal laser scanning microscopy after incubation with U251 cells for 3 h (Figure 2). Free Dbait-YOYO-1 was sparsely located in the cytoplasm. Meanwhile, micelle-treated groups exhibited strong Dbait-YOYO-1 fluorescence, which indicated that Dbait significantly benefited from micellar delivery. Free DOX co-localized with DAPI, indicating that the DOX had entered the nuclei during 3-h incubation. In the micelle groups, Dbait-YOYO-1 and DOX were mainly found in the perinuclear regions and partially localized to the nuclei, suggesting that a longer time interval was required to profusely accumulate in the nuclei. Meanwhile, the more abundant cellular fluorescence intensity of ch-K5(s-s)R8/(Dbait-DOX) group and ch-K5(s-s)R8-An/(Dbait-DOX) + MMP-2 group compared to the ch-K5(s-s)R8-An/(Dbait-DOX) without MMP-2 group confirmed that the R8 moiety promoted micelle internalization.

In vitro micellar cytotoxicity and anti-proliferation effect of collaborative treatment
The toxicity of the different micelle formulations compared to that of free DOX and empty ch-K5(s-s)R8-An micelles was investigated in U251 cells. As shown in Figure 3A, 48-h incubation with empty ch-K5(s-s)R8-An micelles did not produce any toxicity up to a polymer concentration of 30 μg/mL. Therefore, the toxicity observed in cells treated with DOX, ch-K5(s-s)R8-An/DOX micelles, and ch-K5(s-s)R8-An/ (Dbait-DOX) micelles in combination with RT were attributed to the DOX and Dbait/RT regimens and not the polymer carrier ( Figure 3B). The ch-K5(s-s)R8-An/(Dbait-DOX) micelles plus RT exhibited the lowest IC 50 (0.692 μg/mL), which was approximately 2.7-and 1.5-fold lower than that of free DOX and ch-K5(s-s)R8-An/DOX micelle-delivered DOX, respectively ( Figure 3B).

In vitro DNA damage response to combined treatment
RT and DOX are known to produce DNA double-strand breaks (Biau et al., 2019). Dbait promotes phosphorylation of nuclear DNA-PK targets and prevents DNA damage repair proteins from detecting actual chromosome double-strand breaks (Giovannini et al., 2014). Nuclear γ-H 2 AX serves as an indicator of DNA double-strand breaks and will accumulate when the DNA double-strand breaks are not repaired (Kavanagh et al., 2013;. γ-H2AX levels were therefore analyzed by Western blot. As shown in Figure 4B and C, U251 cells treated with ch-K5(s-s)R8-An/(Dbait-DOX) + RT expressed significantly higher levels of γ-H 2 AX compared to those in the ch-K5(s-s) R8-An/Dbait + RT and ch-K5(s-s)R8-An/DOX treatment group. Levels of DNA repair proteins phospho-P53 (p-P53, Figure 4D) and DNA-PKcs ( Figure 4E) were consistent with the extent of DNA double-strand breaks (γ-H 2 AX, Figure  4C) after ch-K5(s-s)R8-An/(Dbait-DOX) + RT treatment. These results suggested that ch-K5(s-s)R8-An/(Dbait-DOX) + RT synergistically improve RT efficacy by raising the DNA damage level and duration in U251 cells.

In vivo brain targeting and biodistribution of ch-K5(s-s)R8-An micelles
Investigation of in vivo biodistribution is essential for the evaluation of the safety and potential efficacy of drug delivery systems (Feng et al., 2014). As illustrated in Figure 5A, no free BODIPY fluorescence (negative control) was detected in the cranial region of glioblastoma-bearing mice during 24 h of circulation, which was confirmed in the postmortem brain ( Figure 5B). In contrast, BODIPY labeled ch-K5(s-s)R8 and ch-K5(s-s)R8-An micelles were observed in the brain area at 4 h after injection and their brain accumulation gradually increased during the subsequent 20 h ( Figure 5A), which lead to high BODIPY fluorescence in the excised brains of mice treated with ch-K5(s-s)R8-An/BODIPY micelles. Confocal laser scanning microscopy of histological sections of excised brains revealed that ch-K5(s-s)R8-An/BODIPY micelles and ch-K5(s-s) R8/BODIPY micelles had accumulated in the tumor site 4 h after intravenous injection ( Figure 5C). Strong BODIPY fluorescence was observed in the tumor tissues with ch-K5(s-s) R8-An/BODIPY micelles, which underscored the brain-targeting utility of angiopep-2 and the in situ glioblastoma-targeting ability of ch-K5(s-s)R8-An micelles.

In vivo anti-glioblastoma efficacy of combined treatment
In vivo anti-tumor effect of different formulations and the synergistic effect of ch-K5(s-s)R8-An/(Dbait-DOX) micelles combined with RT were further investigated from multiple angles in U251 tumor-bearing nude mice.
First, intracranial tumors were imaged and visualized by brain MRI. As shown in Figure 6A, tumors in control mice that had received physiological saline were the largest, while mice treated with DOX alone exerted only a modest inhibitory effect on intracranial tumor growth. In contrast, therapeutic efficacy was observed in the following order: RT < ch-K5(s-s)R8-An/DOX micelles ≈ ch-K5(s-s)R8-An/Dbait micelles + RT < ch-K5(s-s)R8/ (Dbait-DOX) micelles + RT < ch-K5(s-s)R8-An/(Dbait-DOX) micelles + RT. The tumor inhibitory effect was most pronounced in the ch-K5(s-s)R8-An/(Dbait-DOX) micelles + RT group compared to the other groups, highlighting the sensitizing effect of DOX and Dbait combined with RT, which further proved that angiopep-2-modified micelles could enhance the antitumor efficacy due to the brain-targeting attributes in vivo.
Second, therapeutic efficacy was measured by monitoring body weight ( Figure 6B) and Kaplan-Meier survival analysis ( Figure 6C, Table S2), which are negative indicators that are affected by the growth of orthotopic glioblastoma Shan et al., 2016). The body weight of tumor-bearing mice in the control group began to decrease rapidly 2 weeks after tumor cell inoculation. Mice that had been treated with ch-K5(s-s)R8-An/(Dbait-DOX) micelles + RT experienced the longest post-inoculation weight gain and the least weight deterioration. When compared with the saline group, the median survival times of mice treated with free DOX, RT, ch-K5(s-s)R8-An/Dbait + RT, ch-K5(s-s)R8-An/DOX, and ch-K5(s-s)R8-An/(Dbait-DOX) + RT were increased by 23.1%, 26.92%, 51.9%, 55.8% and 69.2%, respectively (Table S2). In contrast, ch-K5(s-s)R8/(Dbait-DOX) + RT combination treatment was the most effective treatment, with a 115.4% median survival extension compared to the saline treatment by increasing the median survival time from 26 days to 56 days. Those results indicated that the ch-K5(s-s)R8-An/ (Dbait-DOX) co-delivery system can synergistically act with RT by suppressing U251 solid tumors and increasing the survival of tumor-bearing mice in vivo.
Third, we assessed histological tissue damage ( Figure 6D) and γ-H 2 AX immunofluorescence ( Figure 6E) in tumor biopsies after the completion of therapeutic regimens. Compared to other groups, mice in the ch-K5(s-s)R8-An/(Dbait-DOX) micelles + RT group exhibited a higher extent of nuclear pyknosis and liquefactive necrosis, with an increased γ-H2AX level in orthotopic tumor sections. These results confirmed the in vivo synergistic cytotoxicity and DNA damaging effect of the chemo-radiotherapy where ch-K5(s-s)R8-An/(Dbait-DOX) micelles were used as the radiosensitizer delivery system.
In addition, histological analysis was performed to examine the in vivo safety of systemic administration of ch-K5(s-s) R8-An/(Dbait-DOX) micelles and RT (Figure 7). In agreement with the reported cardiotoxicity of DOX (Zou et al., 2018), mice in the free DOX group showed cardiac muscle damage in the form of distorted, swollen, denatured myocardial cells and disorganized, fractured myocardial fibers that appeared in a wavy pattern. However, no obvious cardiotoxicity was observed in any of the micelle-treated or RT-treated mice compared to the control group. Also, none of the micellar formulations imparted in vivo toxicity to the brain, liver, spleen, lungs, and kidneys, indicating the in vivo safety of the combination treatment with systemically administered micelles and the locally administered radiation in mice.

Discussion
In this study, we have developed a novel brain-targeting and tumor microenvironment-responsive micelle formulation (ch-K5(s-s)R8-An) for the co-delivery of radiosensitizer Dbait and the chemotherapeutic drug DOX into malignant glioblastoma tumors. This Dbait and DOX co-delivery system also showed high drug loading capacity and good drug release behavior in vitro. The mean diameter of these micelles favors a long circulation time and renders the micelles suitable for systemic administration and tumor targeting via an enhanced permeability and retention effect (Acharya & Sahoo, 2011;Li et al., 2022). Also, the positive zeta potential of ch-K5(s-s)R8-An/(Dbait-DOX) is conducive to electrostatic interactions between the drug delivery system and the negatively charged cancer cell membrane, which may facilitate greater uptake efficiency and accumulation in tumor tissue (Perrault et al., 2009;Kuo et al., 2018). In addition, its environment-dependent release kinetics are favorable for the release of micelle-delivered Dbait and DOX in tumor tissue (You et al., 2021). Based on the results, it is expected that the micelle particles will remain relatively stable and retain Dbait and DOX in plasma. Once delivered into tumor cells, Dbait and DOX are released into the cytosol under the lysosomal acidic environment and cytoplasmic reductive environment (Wang et al., 2015).
In our previous observations (Jiao et al., 2019), ch-K5(s-s) R8-An micelles exit lysosomes by raising the lysosomal pH, leading to osmotic swelling and bursting of lysosomes. Therefore, it is anticipated that the cytoplasmic release and nuclear transport of micellar Dbait and DOX occur after the lysosome escape and disulfide bond dissociation in the ch-K5(s-s)R8-An/(Dbait-DOX) micelles. In this study, ch-K5(s-s) R8-An/(Dbait-DOX) micelles promoted the uptake of Dbait and DOX through a different mechanism than that of free DOX (Lee et al., 2008;Gong et al., 2017), which is consistent with an earlier report that free DOX passively and rapidly diffuses into nuclei after 2-h incubation whilst nanoparticulate DOX is still retained in the cytoplasm (Prabaharan et al., 2009). Furthermore, the ch-K5(s-s)R8-An micelles showed rapid accumulation in the in situ glioblastoma xenografts after intravenous administration, with the most abundant accumulation occurring in the brain region of glioblastoma-bearing mice. This positive brain-targeting property of ch-K5(s-s)R8-An micelles was consistent with our previous report (Jiao et al., 2019), indicating that the angiopep-2 modification effectively enhances the tumor accumulation of ch-K5(s-s)R8-An micelles due to its brain-targeting characteristics.
In addition, the co-delivery system combined with RT treatment exhibited a synergistic anti-glioblastoma effect in U251 cells and in situ U251 tumor xenografts. Even though the DNA damage level ( Figure 4C) was just slightly greater than that induced by ch-K5(s-s)R8-An/Dbait + RT treatment, ch-K5(s-s)R8-An/(Dbait-DOX) + RT treatment still contributed to a more marked cell death rate ( Figure 3C) and apoptotic phenotype ( Figure 4A) at 48 h post-RT. This notable phenomenon suggested that the addition of DOX to Dbait + RT therapy expedites and/or augments the extent of cell death. In line with in vitro results, ch-K5(s-s)R8-An/Dbait + RT treatment yielded the smallest tumor size, strongest tumor tissue injury, and highest DNA damage in orthotopic U251 tumors, which increased the median survival time of U251 tumor-bearing mice from 26 days to 56 days. These results highlighted the complementary therapeutic potency of the combined ch-K5(s-s)R8-An/(Dbait-DOX) plus RT treatment.
In summary, we developed a novel brain-targeting and tumor microenvironment-responsive micelle formulation (ch-K5(s-s)R8-An) for the co-delivery of radiosensitizer Dbait and the chemotherapeutic drug DOX into malignant glioblastoma. This ch-K5(s-s)R8-An co-delivery system showed good glioblastoma targeting specificity and reduced DOX-mediated cardiotoxicity. The combined chemo-radiotherapy of ch-K5(s-s) R8-An/(Dbait-DOX) plus RT exhibited the most remarkable in situ tumor inhibition outcome with smaller tumor size, longer median survival time, and higher DNA damage rate than any other separate treatment. Therefore, this dual-targeting and microenvironment-responsive micellar system shows promise as an advanced co-delivery system for Dbait and hydrophobic drugs such as DOX for the chemo-radiotherapy of malignant glioblastoma.